Increasing numbers of HIV-infected adults in Africa need hospital care. It remains unclear what impact this has on health care services or on how hospitals respond. The aim of this study was to describe the effects of a rising case load of adult HIV-related disease by comparing results from a prospective cross-sectional study of acute adult medical admissions to a government hospital in Nairobi conducted in 1992 with results from a previous study done in 1988 and 1989 in the same hospital, using the same study design and protocol. Data on age, gender, number admitted, length of stay, HIV status, clinical AIDS, final diagnosis, case mix, and outcome were compared. In 1992, 374 consecutive patients were admitted in 15 24-hour periods (24.9 patients/period) compared with the 1988 to 1989 study, which enrolled 506 patients in 22 24-hour periods (23.0 patients/period). Patients' age, gender, and length of hospital stay were similar in both studies. In 1992, 39% of patients were HIV-positive compared with 19% in 1988 to 1989 (p < 10(-6)); whereas seropositive admissions rose 123% between the two periods (p < .0001), HIV-negative admissions declined 18% (p < .05). Clinical surveillance for AIDS consistently identified <40% of HIV-positive patients. Irrespective of HIV status, tuberculosis and pneumococcal pneumonia were the leading diagnoses in both surveys. No change was found in the diagnoses recorded for HIV-positive patients, but in HIV-negative patients, reductions were significant in the case mix (p < .00001) and range of diagnoses (p < .001) seen in 1992. Outcome remained unchanged for HIV-positive patients with approximately 35% mortality in both surveys. Outcome significantly worsened, in relative and absolute terms, for HIV-negative patients: in 1992, mortality was 23%, compared with 13.9% in 1988 to 1989 (p < .005), with 3.5 deaths per 24-hour period in 1992 compared with 2.6 deaths per 24-hour period in 1988 to 1989 (p < .05, one-tailed). These data suggest that increasing selection for admission is taking place as demand for care increases because of HIV/AIDS. This process appears to favor HIV-positive patients at the expense of HIV-negative patients who seem to be crowded out and, once admitted, experience higher mortality rates. The true social costs of the HIV epidemic are underestimated by not including the effects on HIV-negative people.
